HORTSCIENCE 28(11):1129-1130. 1993.

The camera lens, placed 94 cm above the root

system, produced a 15 × 14-cm image that was
Measuring Root Surface Area and digitized to512 × 480 pixels with gray values
ranging from 0 to 255. These images were
either processed immediately or stored for
Mean Root Diameter of Peach future analysis. All image processing involved
the area profile integration software of the
Seedlings by Digital Image Analysis Tracer-Northem model 5502 advanced image
processing system (Tracer-Northem, Mid-
M. Tagliavini1, L.J. Veto, and N.E. Looney dleton, Wis.).
Root surface area was estimated by 1)
Agriculture Canada Research Station, Summerland, BC V0H 1Z0, Canada determining a gray-value threshold corre-
Additional index words. Prunus persica, growth regulators sponding to the finest roots displayed on the
video image; 2) producing a binary image (all
Abstract. Using area profile integration software and an image processing system, we roots are black, and the background—all pix-
reliably estimated total root surface area of intact each [Prunus persica (L.) Batsch.] els with gray values below the threshold-is
seedlings by 1) producing high-quality monochromatic video images under preset and white); 3) determining the total number of
constant conditions; 2) determining a threshold gray intensity value that differentiates the pixels occupied by roots; 4) determining the
finest roots on the image; 3) producing a binary image where all pixels with gray values proportion of the total pixel matrix occupied
above the threshold are black; 4) determining the proportion of black pixels on the 480× by a root image and transforming this to a two-
512-pixel matrix; and 5) multiplying this two-dimensional root surface value by π to dimensional root surface area; and 5) multi-
estimate total root surface area. Normalized intensity (an average intensity weighted plying the two-dimensional root surface area
according to the proportion of the binary image in each gray scale class) was calculated by π to obtain total root surface area (roots are
using software that superimposed the video image on the binary image and was used to assumed to be cylindrical.).
estimate mean root diameter. Evidence of reliability and examples of the use of both Estimating average root diameter. Video
estimates are provided. image analysis can be used to estimate average
root diameter by assuming that thicker roots
Intact root systems are difficult to measure image processing systems with regard to accu- produce a darker gray level on the video image
with a high degree of accuracy. The task is racy and capacity (Baxes, 1984; Green, 1989). than thinner roots and that the entire video
even harder if the aim is to make periodic This paper describes an image analysis image of a root system is composed of pixels
measurements of root growth and root system approach to estimating total surface area and displaying up to 256 gray levels. The
development on the same plant. These diffi- average diameter of white roots and demon- Tracer-Northem software, developed to ana-
culties probably account for our meager un- strates the utility of this nondestructive method lyze fiber thickness for the pulp and paper
derstanding of how root environment affects in experiments with peach seedlings grown in industry, was used to determine average root
total root surface area and mean root diameter, solution culture. We have not tested this diameter. For this specific application, we
which are physiologically important indica- procedure’s utility for estimating root system processed composite samples of peach roots
tors of development. Total root surface area characteristics over times when root suberiza- of known length and diameter (determined
likely influences the kinetics of water and tion and the root system’s physical size could with an ocular micrometer) with the video
mineral uptake (Smika and Klute, 1982), and become limiting factors. image analysis procedure described earlier to
root diameter influences net ion influx into Estimating surface area. Developing this determine root surface area. The black and
roots (Peterson and Barber, 1981). procedure involved capturing monochromatic white binary image (step 2) is then superim-
The traditional methods for measuring root video images of the entire root systems of posed onto the original video image. The ratio
systems have been reviewed by Bohm (1979). peach seedlings grown for 38 days in solution between the digitized root image intensity and
Total root surface area can be determined from culture. The Dage-MTI series 68 video cam- the binary root image area is the value referred
the sum of all root length and diameter mea- era (Dage-MTI, Michigan City, Ind.) was to as “normalized intensity.” The natural loga-
surements, by solution uptake methodology, equipped with a 110-mm macro lens. The rithm (in) of normalized intensity was used to
and more recently, by digital image analysis. seedlings, blotted dry and set between glass estimate average root diameter using the equa-
Mean root diameter can be calculated from plates to give a relatively flat image, were tion Y = 1063X-1.5341, where Y is the mean root
surface area and root length (or root volume) subjected to a preset combination of transmit- diameter, and X is the normalized intensity.
by assuming that roots are cylindrical. ted and reflected light to eliminate shadows. The coefficient of determination (r2) of 0.96
Digital image analysis has several advan-
tages (Stutte, 1990), the most important being
speed and accuracy. It has been used to esti-
mate root surface area of corn (Zea mays L.)
(Smika and Klute, 1982) and onion (Allium
cepa L.) (Ottman and Timm, 1984) grown in
soil or in nutrient solution, and the algorithm
used to calculate the surface area of both
groups has been refined by Yanuka and Elrick
(1985). Recent developments in electronic
hardware and software have greatly improved
Received for publication 23 Mar. 1992. Accepted
for publication 13 June 1993. Summerland Re-
search Station contribution 806. The cost of publish-
ing this paper was defrayed in part by the payment
of page charges. Under postal regulations, this paper
therefore must be hereby marked advertisement
solely to indicate this fact.
1
To whom reprint requests should be addressed.
Current address: Istituto di Coltivazioni Arboree,
Univ. of Bologna, 40126 Bologna, Italy.

HORT SCIENCE , VOL. 28(11), NOVEMBER 1993 1129

 yl] −α− (1,1-dimethyl)-1 H- 1,2,4-triazole-1-
ethanol (paclobutrazol, PBZ)/liter, or a com-
bination of these plant growth regulators in the
nutrient solution. In a second experiment
(Tagliavini et al., 1991), ‘Siberian C’ peach
seedlings grown at an RZT of 16 or 24C
received three levels of P (0.05, 0.5 and 5.0
mM) in the nutrient solution. In both experi-
ments, plants were transported intact to the
image analysis facility, removed from the so-
lution, and blotted dry; root images were ac-
quired and stored in memory; and plants were
placed back into the culture solution without
creating any visible stress or noticeable effect
on subsequent growth. The seedling roots were
out of solution for <2 min.
With both experiments, the procedure de-
tected significant effects of RZT on root-sur-
face area (Table 2; Tagliavini and Looney,
1991). The elevated P level reduced the mean
root diameter of seedlings held at 24C (Table
2), and in the experiment involving plant growth
indicates excellent ability to predict root di- average root diameter × π. regulation (Tagliavini and Looney, 1991), PBZ
ameter from normalized intensity. The stan- Total root surface area as determined by treatment led to a significant increase in mean
dard error of the estimate indicated that a 95% image analysis proved to be highly correlated root diameter (1.13 mm across all RZTs vs.
confidence interval of ±0.04 mm can be ex- to calculated area (Fig. 2), although the slope 0.94 mm for the control).
pected on the average root diameter estimation was <1 (0.90). Thus, the image analysis proce- While our work to date has involved only
(Fig. 1). dure appears to overestimate slightly surface peach seedlings and we have not determined
Finally, to test the utility and reliability of area relative to the calculated value. Similarly, the limits of seedling size or age, we are
both approaches to root system analysis, we we found the two sets of values for root diam- satisfied that these procedures can be used to
processed whole peach seedling root systems eter (Table 1) to be in close accord, with a reliably estimate root surface area and mean
using digital image analysis and then destruc- smaller cv in the image analysis set. root diameter of root systems growing in solu-
tively examined each root system using other We stress that, each time images are ac- tion culture. The method extends the useful-
procedures. Total root length was determined quired, the investigator must use standardized ness of video imagery as a tool to improve our
using the line intersect method (Tennant, 1976) instrument conditions (i.e., light intensity, type knowledge of root system development. Be-
and the Comair Rootlength Scanner (Hawker of illumination, lens opening, magnification, cause this method is nondestructive, it should
de Haviland Victoria, Melbourne, Australia). camera mode, working distance). The proce- prove particularly useful in experiments study-
Average diameter of a subsample of roots dure for estimating mean root diameter re- ing the effects of root-zone temperature, sub-
dispersed on a 1 × 1-cm grid was determined quires using both transmitted and reflected strate fertility, and plant-growth-regulating
with an ocular micrometer. Root surface area light to obtain complete differentiation of root chemicals on the kinetics of root system devel-
was calculated as the product of root length × diameters in terms of pixel intensities. It is opment.
essential to use a new calibration and new
Literature Cited
equations when standard conditions change.
We also obtained very satisfactory results Baxes, G.A. 1984. Digital image processing, a practical
when we used the image analysis approach to primer. Prentice Hall, Englewood Cliffs, N.J.
nondestructively estimate root characteristics Bohm, W. 1979. Methods of studying root systems.
in two other experiments involving peach seed- Springer-Verlag, New York.
Green, W.B. 1989. Digital image processing: A sys-
lings grown in nutrient solution. In the first tems approach. 2nd ed. Van Nostrand Reinhold,
(Tagliavini and Looney, 1991), ‘Rutger’s Red New York.
Lear peach seedlings were exposed to a root- Ottman, M.J. and H. Timm. 1984. Measurements of
zone temperature (RZT) of 10, 16, or 22C and viable plant roots with the image analyzing com-
were subjected to 0 or 2 mg gibberellic acid puter. Agron. J. 76:1018–1020.
Peterson, W.R. and S.A. Barber. 1981. Soybean root
(GA3)/liter, 0 or 2 g β− [(4-chlorophenyl)meth- morphology and K uptake. Agron. J. 73:316-319.
Smika, D.E. and A. Klute. 1982. Surface area measure-
ment of corn root systems. Agron. J. 74:1091–
1093.
Stutte, G.W. 1990. Analysis of video images using an
interactive image capture and analysis system.
HortScience 25:695-697.
Tagliavini, M., E.J. Hogue, and G.H. Neilsen. 1991.
Influence of phosphorus nutrition and root zone
temperature on growth and mineral uptake of peach
seedlings. J. Plant Nutr. 14:1267–1276.
Tagliavini, M. and N.E. Looney. 1991. Response of
peach seedlings to root-zone temperature and
root-applied growth regulators. HortScience
26:870-872.
Tennant, D. 1976. A test of a modified line intersect
method of estimating root length. J. Ecol. 63:995–
1001.
Yanuka, M. and D.E. Elrick. 1985. Applications of
microcomputer-based image digitization in soil and
crop sciences. Computers & Electronics Agr. 1:59–
73.

1130 HORT SCIENCE , VOL. 28(11), NOVEMBER 1993